Towards High Accuracy Pedestrian Detection on Edge GPUs

Abstract

Despite the rapid development of pedestrian detection algorithms, the balance between detection accuracy and efficiency is still far from being achieved due to edge GPUs (low computing power) limiting the parameters of the model. To address this issue, we propose the YOLOv4-TP-Tiny based on the YOLOv4 model, which mainly includes two modules, two-dimensional attention (TA) and pedestrian-based feature extraction (PFM). First, we integrate the TA mechanism into the backbone network, which increases the attention of the network to the visible area of pedestrians and improves the accuracy of pedestrian detection. Then, the PFM is used to replace the original spatial pyramid pooling (SPP) structure in the YOLOv4 to obtain the YOLOv4-TP algorithm, which can adapt to different sizes of people to obtain higher detection accuracy. To maintain detection speed, we replaced the normal convolution with a ghost network with a TA mechanism, resulting in more feature maps with fewer parameters. We constructed a one-way multi-scale feature fusion structure to replace the down-sampling process, thereby reducing network parameters to obtain the YOLOv4-TP-Tiny model. The experimental results show that the YOLOv4-TP-tiny has 58.3% AP and 31 FPS in the winder person pedestrian dataset. With the same hardware conditions and dataset, the AP of the YOLOv4-tiny is 55.9%, and the FPS is 29.

Exploring the Practice of Teaching English Professional Translation Using IoT and Cloud Computing Multimedia Assistance

“…Multimedia-assisted professional translation in English language teaching has certain advantages. First of all, it enriches the teaching content [4]. At present, the teaching equipment of English simultaneous interpretation in colleges and universities is basically perfect, and the classroom multimedia is getting richer.…”

Exploration of Exercise Characteristics and Psychological Impact of Cold Urban Elderly Based on Multivariate Data Analysis

“…Cold cities are characterized by low average annual temperatures, obvious seasonal changes, long winters, less sunshine, and shorter days [1]. In winter, low temperature, snow, freezing, and other climatic characteristics will cause a lot of inconvenience for the elderly to travel and exercise, and most of the outdoor fitness facilities are difficult to use, which have a greater impact on the outdoor fitness exercise of the elderly and restrictions [2][3][4][5][6]. Therefore, the elderly population in cold cities has a greater demand for indoor sports and fitness, especially in winter, and active participation in indoor fitness can effectively compensate for the lack of outdoor sports and create more favorable conditions for the elderly to enjoy a colorful sports life [7].…”

Marketing Strategy of Rural Home Stay Industry Based on Intelligent IOT

“…The sinking market has huge space, with 300 prefecture-level cities, 2,800 counties, 40,000 towns, 660,000 villages, and about 1 billion people [21][22][23]. In addition, the national top-level policy supports the promotion of rural revitalization and encourages the development of farm stays and leisure farm tourism, further benefiting the B&B industry [24][25][26][27]. Vigorously promote the construction of digital countryside, focus on solving practical problems, and expand the application scenarios of agricultural and rural big data [28].…”